- Sources: incident, HN discussion
- Summary: GitHub opened the incident at 09:16 UTC on 2026-09-13 for degraded availability of API Requests, Issues, Pages and Pull Requests, and added Actions at 09:25 UTC. At 09:36 UTC it attributed the failure to increased database replication delays on the collab cluster, which raised error rates in authorization endpoints and produced follow-on errors across the system. GitHub reduced load with internal load-shedding, reported signs of recovery at 10:26 UTC, and resolved the incident at 10:44 UTC.
- Why it matters: One cluster's replication delay took out API requests, Issues, Pull Requests, Actions and Pages together, because the authorization path they share failed first.
- Follow-up: The detailed root cause analysis GitHub states will follow.
send feedback on this story